Even if each sendCAS call is made from the same thread you can use a CAS pool so that multiple CASes are in process at the same time. You should make the size of the pool at least as large as the number of analysis engine threads deployed. If it is larger then CASes will be queued on the endpoint until the next analysis engine is ready for work.
Burn On Fri, Dec 22, 2017 at 10:43 AM, Benedict Holland < [email protected]> wrote: > I am very interested in this as well. > > Do you happen to have N threads running in a pool and N CAS instantiated > objects? I admit though, I have very little experience with the Async > engine. > > Thanks, > ~Ben > > On Fri, Dec 22, 2017 at 8:23 AM, Matthias Koch <[email protected]> > wrote: > > > Hi > > > > We noticed that UIMA AS does not scale with multiple threads as expected. > > > > Digging further into the problem we noticed two places which are executed > > by at most one thread at a time: > > - BaseUIMAAsynchronousEngineCommon_impl.sendCAS() > > - BaseMessageSender.run() > > We are using one UimaAsynchronousEngine per endpointName, thus > serializing > > and sending a message will only be done one at a time, > > no matter how many instances of this analysisEngines we have running > under > > this endpointName. > > > > How can we improve on this? > > Can we just pool the UimaAsynchronousEngines? > > Can we do better than just calling UimaAsynchronousEngine.sendAnd > > ReceiveCAS()? > > > > Sincerely, > > Matthias > > > > -- > > Matthias Koch > > > > Averbis GmbH > > Tennenbacher Str. 11 > > 79106 Freiburg > > Germany > > > > Fon: +49 761 708 394 0 > > Fax: +49 761 708 394 10 > > Email: [email protected] > > Web: https://averbis.com > > > > Headquarters: Freiburg im Breisgau > > Register Court: Amtsgericht Freiburg im Breisgau, HRB 701080 > > Managing Directors: Dr. med. Philipp Daumke, Dr. Kornél Markó > > > > >
